-
I have Calamari 2.2.1 / Python 3.7.4 / TF 2.4.1 / Numpy 1.21.6, and attempt calamari-train in a (Power9-based) HPC environment.
Unfortunately, I get the following exception:
stacktrace
```
…
-
Trying not to use a recurrent LSTM network for avoiding the issue in deserialization (issue #72) I am trying to train the model without a recurrent LSTM layer, using the "finest" training method. I ge…
-
I am not able to read a model containing a LSTM layer from an onnx file, saved after training the model.
Settings:
```
Python 3.8.6 | packaged by conda-forge | (default, Oct 7 2020, 19:08:05)
…
-
This template is for miscellaneous issues not covered by the other issue categories.
For questions on how to work with TensorFlow, or support for problems that are
Epoch 1/2
---…
-
I have a network that works fine on iPhone 6s with iOS 12, iPhone 7 with iOS 11 and on iPhone 6s+ with iOS 11 both on GPU as well as on CPU.
However on iPhone X with iOS 12 the network predicts NaN…
-
`stateful` LSTMs seem to be confusing everybody. I don't recommend `stateful` unless you know what it is and have a good reason to use it.
Imagine you have a 2 layer neural network and you only tra…
-
In an HPC environment (where I want to `calamari-train`), with Power9 CPU (`ppc64le` architecture), I get
```
pip install tensorflow-addons
ERROR: Could not find a version that satisfies the requ…
-
- would be great to get quick eyes from @buechler314 and @anyaelena
- still in the investigation phase
- so far the symptom is like:
- **training is successfully done when CPU is only requ…
-
_I posted this bug report on the PyEDDL section. I file a new issue here since the bug has been reproduced also in C++._
The C++ code is available [here](https://github.com/deephealthproject/pyeddl…
-
keras:
Layer (type) Output Shape Param # Connected to
==========================================================================================…
ghost updated
2 years ago